Extreme ISO

This past February I began prep on another Lifetime Original film, “BORDERLINE MURDER”, written and directed by Andrew C. Erin.  From the beginning I knew we would have less time and less money than on our previous film together “Final Sale”.  With an extremely short  shooting schedule of 16 days we knew we had to adjust our approach to the film.  While our previous film together had been a slick urban crime drama set in downtown Los Angeles, BORDERLINE MURDER, on the other hand set in a small Texas boarder town would be gritty, raw and less polished.  

Our first major choice that dictated the style of the film was the decision to go completely handheld, allowing us the flexibility to move incredibly fast.  The next thing we did was severaly downsize our Lighting and Grip packages.  Where on the previous film with Andrew we carried a 10-Ton electric and 5-Ton grip, for BORDERLINE MURDER, I limited myself to a shared Grip/ Eletric 3-Ton.  The third and most crucial decision came about after I had heard of people using the MXs extreme high ISOs of 1600 and 2000 in order to increase the overall dynamic range of the image.  I was extremely skeptical at first and decided that before production it would be essential to shoot camera tests.  I’m a firm believer in shooting tests before each project, even if you’re using the same camera and same lenses from project to project.  Every script is different and there is always a different look or image that one is trying to achieve and the process of shooting tests with a particular project in mind I have found to be invaluable.  I’m constantly learning how far I can pull and push a film stock, or in the case of Red, the physical sensor itself.

Like our shooting schedule, our prep time was limited as well.  While I was busy prepping and storyboarding with the director, I was able to have my 1st AC Parker Tolifson shoot a multitude of ISO tests under different conditions at Sim Video, our camera house.  When I saw the results of the test the next morning at the edit suite I was more than pleasantly surprised.  We did tests from 500 ISO -  2000 ISO.  Between 500 - 1200 ISO, I saw what one would expect to see, around 3 stops over key you began to loose all information, this being the typical dynamic range of the Red MX.  Where it got interesting was once you started exposing for 1600 up to 2000, I was literally able to pull detail out of highlights that were 4 stops over key, an “apparent” increase in the total dynamic range (latitude) of the image.  It is what I had heard rumors and whispers of, but here I was sitting in an edit suite and seeing it for myself.  Of course with an extreme ISO of 1600 to 2000 there is an inherent noise floor in the quality of the image, but remember I was after a gritty, grainy and raw look.  The threshold for what is an acceptable noise floor is entirely an asthetic choice, the same way filmmakers choose to shoot on 16mm over 35mm regardless of budget contraints.

But what about the argument that while you can rate the camera as fast as 1600, you’re still only recording the native raw ISO of the MX, which is generally accepted as 500.  So are you really increasing the dynamic range or the ISO of the camera?  One could argue that all you’re doing is applying a LUT of 1600 in camera as oppose to applying it later in post, since regardless the camera’s native ISO is 500.  This is entirely correct, rating the camera at 1600 ISO and viewing with Red Color on-set all you’re doing is applying metadata that will be later used to adjust the raw image in the final grade.  So what’s the point?

By rating at 1600 and making sure your highlights stay within a 4 stop range over key, what in essence you’re doing is by and large extremely under-exposing your raw negative (500 ISO).  There would be times onset where we would switch between RAW and Red Color and be amazed at how dark the RAW image would be.  Dark as in pitch black, then I would switch back to RED COLOR and see the desired image.  I was blown away by how well the low end of the MX chip holds up.  In conclusion, being able to severaly underexpose the raw negative onset enabled an overall increase in the dynamic range of the image, when the desired LUT was applied in post.

Post production was extremely skeptical at first, once they started getting dailies in at first I think there was a bit of a panic, there would be sequences that looked completely blown out white when transcoded traditionally, however when they dialed the ISO back to 1000 as oppose to 1600 all the detail and information was there.  They’ve since locked picture and we’re in the beginning stages of coloring on BORDERLINE MURDER, and I’m a bit surpised myself at how healthy and strong our negative image we have to work with is.  I’ve never had a digital project that had retained all it’s detail from the high end to the low end through out, especially with such a minial lighting and grip package and the speed at which we were moving on set. Shooting 6-8 pages a day and averaging 55-65 setups a day with two cameras.

Because I had such a minimal lighting package, consisting bascially of (2) 1.2K HMIs, (1) 4 K HMI and a Nine light Faye, I was able use the 4K how I would have used a 12K, due to needing infinately less light.  At times I would walk on set and be amazed how “dark” it physically was in the room, but once I got the camera on my shoulder and looked through the viewfinder I was reminded of how fast an ISO of 1600 really is. Where as typciall I would be forced to go direct with a 4K, just to try to get the stop needed in a scene, with rating at 1600 I was able to bounce and even double and tripple difuse the source allowing for a much “bigger” feel to the lighting than one would expect out of such a small lighting package.

It took a while for everyone, including my Gaffer and 1st AC to get on the same page with how I was exposing the film.  I think the first week people though I was crazy, rating day exteriors on a Red MX a 1600 and putting so much ND in front of the lens that it’s the one case that you can justify having an electronic view finder over an optical one.  I was routinely using ND 1.5 and a True Pola on my day exteriors, sometimes even using as much as ND.6 on day interior scenes.  From my tests in prep I knew we’d be able to get back the 4 stops of overexposing it wasn’t untill after week one I was able to go in with the fooatge and look at it through RedCine X that I truely knew we were onto something.  I was amazed how the raw looked when combined with Red’s new RedLogFilm LUT, it was a super smooth curve I’d only been use to seeing out of an Alexa or a Genesis.

While shooting at an extremely high ISO 1600 - 2000 on the Red MX might not be for everyone, it has truely changed the way I approach and expose the MX sensor.  I would have never thought I’d be able to hold 4 stops of highlight detail on a Red.  I can honestly say It’s not a myth.  

BORDERLINE MURDER airs on LIFETIME, July 25th and July 26th.

Written by John Matysiak